Download Low-Noise 0.8–0.96- and 0.96–1.12-THz Superconductor–Insulator

IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 2, FEBRUARY 2006
547
Low-Noise 0.8–0.96- and 0.96–1.12-THz
Superconductor–Insulator–Superconductor
Mixers for the Herschel Space Observatory
Brian D. Jackson, Gert de Lange, Tony Zijlstra, Matthias Kroug, Jacob W. Kooi, Jeffrey A. Stern, and
Teun M. Klapwijk
Abstract—Heterodyne mixers incorporating Nb SIS junctions
and NbTiN–SiO2 –Al microstrip tuning circuits offer the lowest
reported receiver noise temperatures to date in the 0.8–0.96- and
0.96–1.12-THz frequency bands. In particular, improvements in
the quality of the NbTiN ground plane of the SIS devices’ on-chip
microstrip tuning circuits have yielded significant improvements
in the sensitivity of the 0.96–1.12-THz mixers relative to previously
presented results. Additionally, an optimized RF design incorporating a reduced-height waveguide and suspended stripline RF
choke filter offers significantly larger operating bandwidths than
were obtained with mixers that incorporated full-height waveguides near 1 THz. Finally, the impact of junction current density
and quality on the performance of the 0.8–0.96-THz mixers is
discussed and compared with measured mixer sensitivities, as are
the relative sensitivities of the 0.8–0.96- and 0.96–1.12-THz mixers.
Index Terms—Astronomical satellites, niobium, niobium compounds, radio astronomy, submillimeter-wave mixers, superconductor–insulator–superconductor (SIS) mixers.
I. INTRODUCTION
HE heterodyne instrument for the far-infrared (HIFI) [1],1
is a high-sensitivity, high-resolution heterodyne spectrometer that is being built for the European Space Agency’s Herschel Space Observatory [2].2 The instrument’s 0.48–1.25- and
1.41–1.91-THz frequency coverage will offer astronomers an
unprecedented opportunity to observe a significant fraction of
T
Manuscript received February 4, 2005; revised July 13, 2005. This work was
supported in part by the Technologiestichting STW (of The Netherlands), by the
Nederlandse Organizatie voor Wetenschappelijk Onderzoek, and by the European Space Agency under the European Space Research and Technology Centre
Research Contract 11653/95.
B. D. Jackson is with the SRON National Institute for Space Research,
9747 AD Groningen, The Netherlands and also with the Kavli Institute of
Nanoscience, Faculty of Applied Sciences, Delft University of Technology,
2628 CJ Delft, The Netherlands (e-mail: [email protected]).
G. de Lange is with the SRON National Institute for Space Research, 9747
AD Groningen, The Netherlands (e-mail: [email protected]).
T. Zijlstra, M. Kroug, and T. M. Klapwijk are with the Faculty of
Applied Sciences, Kavli Institute of Nanoscience, Delft University of Technology, 2628 CJ Delft, The Netherlands (e-mail: [email protected];
[email protected]; [email protected]).
J. W. Kooi is with the California Institute of Technology, Pasadena, CA 91125
USA (e-mail: [email protected]).
J. A. Stern is with the Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109 USA (e-mail: [email protected]).
Digital Object Identifier 10.1109/TMTT.2005.862717
1[Online].
2[Online].
Available: www.sron.nl/divisions/lea/hifi
Available: http://www.rssd.esa.int/herschel
the submillimeter and far-infrared spectrum, much of which is
not observable from ground-based telescopes due to absorption
by atmospheric water vapor. In order to take full advantage of
this space-based observatory, the HIFI instrument will incorporate state-of-the-art cryogenic heterodyne mixers, with five superconductor–insulator–superconductor (SIS) mixers covering
the 0.48–1.25-THz band and two hot-electron bolometer (HEB)
mixers covering 1.41–1.91 THz.
Unfortunately, although traditional Nb SIS mixers offer
quantum-limited sensitivities below 0.7 THz [3]–[5], increasing resistive losses in their Nb-based microstrip transmission lines at frequencies above the “gap frequency” of
THz) [6] cause the
Niobium (
sensitivities of these mixers to drop significantly above
0.7 THz [7]. However, previous work has shown that the use of
NbTiN-based3 microstrip RF tuning circuits allows the sensitivity of 0.75–1.0-THz SIS mixers to be significantly improved
[8]–[10]. Furthermore, as is discussed in more detail in [11],
the integration of Nb SIS junctions with a NbTiN–SiO –Al
microstrip tuning circuit in which the NbTiN ground plane
is deposited at 400 C (in place of the room-temperature-deposited films used previously) allows low-noise SIS mixer
operation to be extended to at least 1.12 THz.
Following from these advances, this paper describes the RF
design and performance of the 0.8–0.96- and 0.96–1.12-THz
mixers that have been developed for use in Bands 3 and 4 of
the HIFI instrument. In particular, this paper discusses the integrated designs of reduced-height waveguide embedding circuits and two-junction SIS tuning circuits that yield efficient
coupling of radiation to the mixers’ SIS junctions over each frequency band. Additionally, the impact of the current densities
and leakage currents of the SIS junctions on their mixing performance is discussed in light of simulations of the heterodyne
performance of the mixers’ two-junction tuning circuits. This
mixer model is also used to compare the relative performance
of the 0.8–0.96- and 0.96–1.12-THz mixers.
The optomechanical designs of these mixers (including their
corrugated horns), in addition to the designs of their intermediate frequency (IF) output coupling circuits and their shielding
from external electromagnetic interference, are described
elsewhere [12].
3For simplicity, the compound Nb
Ti N
is referred to as NbTiN throughout this paper.
0018-9480/$20.00 © 2006 IEEE
, where x = 0:3 and 0,
548
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 2, FEBRUARY 2006
II. MIXER AND SIS DEVICE DESIGN
Previous reports of the development of NbTiN-based SIS
mixers made use of two basic mixer geometries: a quasi-optical
mixer incorporating a 0.95-THz twin-slot antenna (see [9])
and a waveguide mixer incorporating a full-height 1-THz
waveguide (see [10]). These results demonstrated that a
NbTiN–SiO –Al microstrip RF matching network can be integrated with “standard” 1- m Nb–Al-AlO –Nb SIS junctions
to yield low receiver noise temperatures up to 1 THz (and [11]
demonstrates that this region of low-noise operation may be
extended to at least 1.12 THz by the use of an NbTiN ground
plane that is deposited at 400 C). However, the fixed-tuned RF
bandwidth of the 1-THz waveguide mixer in [10] was limited
to 100 GHz, whereas the quasi-optical mixer in [10] offered
fixed-tuned bandwidths of 200 GHz. This limitation of the
previously demonstrated waveguide mixer has been addressed
by a redesign of the waveguide embedding geometry [13] and
the SIS device’s on-chip microstrip tuning circuit to optimize
the coupling to the SIS junctions.
The starting point of this redesign (see [14]) was a move
from the full-height 1-THz waveguide geometry that was
used in [10] to scaled versions of the 0.65-THz half-height
waveguide geometry that was used in mixers produced for the
James Clerk Maxwell Telescope in Hawaii [15], [16]. In particular, for each of these designs, Fig. 1(a) presents the effective
source impedance at the input to the on-chip microstrip tuning
circuit that is produced by a combination of the waveguide, the
fixed-depth waveguide backshort, the fused quartz substrate
in the substrate channel (including the RF choke filter on the
substrate), and the “across-the-waveguide” bowtie probe. (This
“source impedance” has been calculated in a three-dimensional
(3-D) electromagnetic field simulator.4) From this plot, it is
seen that the original 1-THz waveguide design is characterized
by a source impedance with a strong frequency dependence
and a large imaginary component. In comparison, the source
impedances of the 0.65- and 0.88-THz half-height waveguide
designs are much less frequency dependent (although they
retain significant reactive components). These results are consistent with those obtained previously in [17] and [18], both in
theoretical calculations and in experiment.
As was also demonstrated in [18], further “improvements”
in the source impedance offered by the waveguide embedding
design are obtained by suspending the SIS device substrate in
the substrate channel (and fine-tuning the dimensions in the RF
choke filter to recenter the passband of the filter on the center
frequency of the design). As is seen in Fig. 1(b), this further
reduces the frequency dependence and reactance of the source
impedance at the input of the on-chip tuning circuit. (For clarity,
only the results for the 0.88-THz design are shown here—the
1.04-THz design is a scaled version of the 0.88-THz design.)
Fig. 2 illustrates the geometry of the 0.88-THz suspended
substrate waveguide geometry that is used in the devices
described here. The critical waveguide and device substrate
dimensions in this design are summarized in Table I. Three-dimensional electromagnetic simulations of the embedding
impedance offered by the waveguide geometries are calculated
4Ansoft
Corporation, Pittsburgh, PA.
Fig. 1. Effective source impedance at the input to the on-chip microstrip
tuning circuit (at the center of the “across-the-waveguide” bowtie probe,
given the fixed-depth waveguide backshort, the fused quartz substrate in
the substrate channel, and the RF choke filter patterned on the substrate
surface). (a) Comparison of the full-height 1-THz waveguide geometry and
the half-height 0.65-THz waveguide geometries used previously with the
“unsuspended” half-height 0.88-THz design. (b) Comparison of half-height
0.88-THz waveguide geometries with suspended and unsuspended quartz
substrates in the substrate channel. The geometries of the mixers analyzed here
are summarized in Fig. 2 and Table I.
for a reference plane at the center of the bowtie waveguide
probe (i.e., at the input to the on-chip microstrip tuning circuit).
The 1.04-THz mixer design that is used here is a scaled version
of the 0.88-THz design.
Finally, the RF designs of the mixers were completed by optimizing the geometry of the twin-junction tuning circuit used in
[10] (see Fig. 3) to maximize the coupling of incident RF power
to the SIS junctions over the 0.8–0.96 and 0.96–1.12-THz
bands, given the frequency-dependent source impedances in
Fig. 1. The resulting coupling to the SIS junctions is plotted
in Fig. 4 for several combinations of waveguide embedding
geometry and SIS device parameters (which are summarized in
Tables I and II, respectively).
A number of features are clearly identifiable in these results. First, it is seen that, with a moderate junction current
kA/cm ), the “suspended” and “unsuspended”
density (
0.88-THz designs offer similar coupling efficiencies (50%–60%
across the 0.8–0.96-THz band, with the “unsuspended” design actually offering slightly higher efficiencies). Moving
kA/cm ) significantly
to a higher current density (
improves the coupling to the junctions (due to the junctions’
product), and the coupling obtained with the
reduced
“suspended” design is slightly better than that obtained with the
JACKSON et al.: LOW-NOISE 0.8–0.96- AND 0.96–1.12-THz SIS MIXERS FOR HERSCHEL SPACE OBSERVATORY
549
Fig. 2. Suspended substrate waveguide geometry incorporating a half-height
waveguide with a bowtie waveguide probe located at the center of the waveguide
and a four-section RF choke filter in the substrate channel. The dimensions
of the waveguide, substrate channel, and substrate are summarized in Table I.
The reference plane at which the embedding impedances are calculated for this
structure is located at the center of the bowtie waveguide probe. In other words,
this reference plane is located at the input of the on-chip microstrip tuning circuit
that is seen in Fig. 3. Ground and dc/IF connections to the SIS device are made
by wire-bonding to gold pads at the two ends of the device (beyond the left and
right edges of this figure).
TABLE I
WAVEGUIDE MIXER EMBEDDING GEOMETRIES ANALYZED IN FIG. 1
Fig. 3. Twin-junction SIS tuning circuit geometry. Top: photograph of a 1-THz
mixer device (adapted from [10]). Note that the dimensions of the microstrip
tuning circuit in this image are slightly different than those in the devices used
here, but that the basic geometry is the same. Bottom: schematic cross section
of the tuning circuit layer structure (adapted from [9]). The dimensions of the
microstrip tuning circuit are summarized in Table II.
tunnel junctions with
m . The microstrip transmission
lines combine a 300-nm NbTiN ground plane, a 250-nm SiO
dielectric layer (with a nominal
), and a 400-nm
Al wiring layer (with a low-temperature dc resistivity of
–
cm).
The fabrication and performance of these mixers are described and discussed further in the following sections.
III. SIS DEVICE FABRICATION
“unsuspended” design. Beyond this, Fig. 4(b) demonstrates that
the coupling to the junctions in the 1.04-THz design is strongly
dependent upon the properties of the NbTiN ground plane—the
use of a film deposited at room temperature is expected to result
in a strong drop in coupling above 1 THz (see [9] and [10]),
whereas the use of a film deposited at 400 C should offer
strong coupling over the full 0.96–1.12-THz band (see [11]).
Based upon these calculations, SIS mixers with three combinations of junction current density, NbTiN quality (supercon, and low-temperature
ducting transition temperature,
normal-state resistivity,
), and embedding geometry
have been produced. Two 0.88-THz mixers (incorporating
and
kA/cm
K,
junctions with
cm and a suspended embedding geometry)
have been produced for Band 3 of the HIFI instrument. Similarly, a 1.04-THz mixer incorporating a suspended-substrate
kA/cm
K,
waveguide geometry with
cm has been produced for Band 4 of
and
the HIFI. In all cases, the SIS junctions are Nb/Al-AlO /Nb
The SIS devices used here were produced using a process
that is derived from that which was used for the demonstrations of quasi-optical and waveguide mixers incorporating
NbTiN/SiO /Al tuning circuits (see [9] and [10], respectively).
However, a number of significant modifications to the process
have been made in order to improve the patterning of the
SIS junctions and the Al wiring layer, and to incorporate
higher-quality NbTiN ground planes in the 1.04-THz mixers.
The primary modifications to the junction definition process
gas mixture to SF
have been: 1) to move from a CF
for the reactive ion etching of the Nb junction electrodes (to improve the anisotropy and repeatability of the etch) and 2) to add
an O plasma etch of the resist pattern following the etch of
the top Nb junction electrode (prior to Ar sputter etching the
Al-AlO barrier and reactive ion etching the bottom Nb electrode). As is discussed in [11], this “resist recessing” step yields
a stepped junction profile in which the edges of the active portion of the Al-AlO barrier are not exposed to the Ar sputter etch
of the barrier, since the final size of the top electrode is reached
550
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 2, FEBRUARY 2006
at the completion of the bottom electrode etch. This is expected
to improve the quality of a typical junction (i.e., to reduce its
leakage current) by reducing the risk of damage to the tunnel
barrier during the etch process.
On top of these changes to the junction definition step, the
other significant changes to the SIS device process have been:
1) using a chlorine-based reactive ion etch to pattern the Al
wiring layer (which offers improved dimension control relative
to the lift-off process that was used in [9] and [10]) and 2) using
NbTiN ground planes deposited at 400 C (at the Jet Propulsion
Laboratory [19]) in the 1.04-THz mixers. Finally, because the
lift-off process that was used previously cannot be used to pattern ground planes that are deposited at high temperatures, these
O .
films were patterned by reactive ion etching in SF
As in [9] and [10], contact UV lithography is used for all resist
pattern definition, RF magnetron sputtering is used to deposit
the SiO dielectric and passivation layers, and dc magnetron
sputtering is used to deposit the Nb, NbTiN, and Al layers.
Table II summarizes the material characteristics and tuning
circuit geometries of the SIS devices that are discussed in the
Sections that follow.
IV. DC CURRENT–VOLTAGE CHARACTERISTICS
Fig. 4. (a) Calculated coupling to the SIS junctions for twin-junction SIS
devices mounted in half-height 0.88-THz waveguides with suspended and
unsuspended substrates. A slight benefit is obtained from the suspended
substrate design if high current-density junctions are used (the values of J in
the legend are given in kA/cm ). (b) Calculated coupling to suspended-substrate
twin-junction SIS devices mounted in a half-height 1.04-THz waveguide.
A clear dependence of the RF coupling on the superconducting properties of
the NbTiN ground plane is observed. Note that the values given the legends
are the “gap frequency” and low-temperature normal-state resistivity of NbTiN
in-THz and 1 cm, respectively F
=h .
(
= 21
)
TABLE II
SIS JUNCTION AND NbTiN GROUND-PLANE CHARACTERISTICS AND TUNING
CIRCUIT GEOMETRIES OF THE SIS DEVICES DISCUSSED HERE
Fig. 5 presents the bias current and IF output power as a
function of bias voltage for two SIS devices (one 0.88-THz
kA/cm and one 1.04-THz device with
device with
kA/cm ). In general, the junction qualities of these devices, as measured by their subgap to normal-state resistance ra, are excellent, with
–
tios
for devices with
kA/cm and
for devices with
kA/cm (at a mixer temperature of
2–2.5 K).
Beyond this, two other features are noted in Fig. 5. First,
the photon step in the 1.04-THz device pumped at 1.14 THz
is barely wide enough to yield a bias point that is not affected
by the Shapiro effect (which can cause instabilities in the IF
output power in the output power peaks seen on either side of
mV/THz). Additionally, whereas
the 0.88-THz devices have a typical series resistance in their Al
wiring layer of 0.7 (which is consistent with the low-temperature dc resistivity of the Al film), the 1.04-THz devices have
a series resistance of 1.5 . Note that the resistivity of the Al
layer was determined from the resistance of long, narrow strips
of Al deposited under the same conditions as the Al in the actual devices, while the the series resistance in the Al wiring
of actual junctions is determined from the slope in the junctions’ zero-voltage supercurrents. The higher series resistance
in the 1.04-THz devices indicates that the resistivity of the Al
wiring layer is higher in the 1.04-THz devices than the
cm that is assumed in the design. This is not fully understood, but is thought to be related to the fact that the surface of
the SiO dielectric layer on which the Al is deposited is significantly rougher in the 1.04-THz devices than in the 0.88-THz devices (due to differences in the growth mechanics of SiO layers
on NbTiN films deposited at 400 C and room temperature).
JACKSON et al.: LOW-NOISE 0.8–0.96- AND 0.96–1.12-THz SIS MIXERS FOR HERSCHEL SPACE OBSERVATORY
551
Fig. 6. Schematic representation of the mixer test receiver. This includes a
vacuum hot–cold load that is used to improve the accuracy of measurements
of the mixer sensitivity by removing the vacuum window from the optical path
to the liquid-nitrogen-cooled cold load, and by eliminating the influence of
atmospheric absorption of submillimeter radiation on the noise measurements.
Fig. 5. (a) Bias current and mixer IF output power as a function of bias voltage
for a 0.88-THz SIS tunnel junction with a current density of J = 13 kA/cm ,
operated at 0.86 THz. (b) and (c) Bias current and mixer IF output power as a
function of bias voltage for a 1.04-THz SIS junction with J = 6:5 kA/cm ,
operated at 1.02 THz [in (b)] and 1.14 THz [in (c)]. The limited bias range that
remains in the 1.04-THz device pumped at 1.14 THz is noted—this is close
to the maximum operating frequency for a “traditional” Nb/Al-AlO /Nb SIS
junction. In both cases, the junction qualities (as measured by their subgap
=R
= 30–60
to normal-state resistance ratios) are excellent, with R
and 20 obtained for J = 6:5 and 13 kA/cm , respectively. (Note that (b) is
adapted from [11].) The vertical markers on the hot and cold IF output power
characteristics in each plot indicate the bias voltage that is typically used for
RF sensitivity testing—(a) indicates the bias voltage that is used for 0.88-THz
mixers, while (b) and (c) indicate the bias voltages that are used for 1.04-THz
mixers below 1.04 and above 1.14 THz, respectively (no LO is available in the
gap between these ranges).
V. RF MEASUREMENT SETUP
The heterodyne sensitivities of the mixers described here have
been obtained from conventional -factor measurements using
a room-temperature “hot” blackbody signal source and a 77 K
“cold” blackbody source (using the Callen–Welton formulation
[20] to determine their effective source temperatures from their
physical temperatures). Furthermore, because this work is focused on the development of mixers for the HIFI instrument
(which includes a window-free, all-reflective optical design), an
attempt has been made to minimize the receiver’s input coupling
losses by: 1) replacing the dielectric focusing lens used previously with a Au-coated mirror on the 4 K stage of the liquidhelium-cooled test cryostat and 2) making use of a vacuum
hot–cold load. In particular, using this vacuum hot–cold load
removes the cryostat’s vacuum window from the optical path
between the mixer and the cold load and eliminates the effect
of absorption by atmospheric water vapor, which can be significant at submillimeter wavelengths. Fig. 6 presents a schematic
representation of the mixer test system, including the vacuum
hot–cold load, in which a rotating mirror located inside the cryostat is used to chop between the hot load (a room-temperature
absorber located outside the cryostat) and the cold load. This
cold load is produced by coating the bottom of a liquid nitrogen
vessel in the receiver’s vacuum system with a mixture of silicon carbide grains in black Stycast epoxy [21]. Previous measurements [22] showed that this coating has an emissivity of
at submillimeter wavelengths.
Differences between the mixer test system and the environment of the HIFI instrument have been further minimized by
the use of a cryogenic intermediate frequency (IF) amplification
system that includes prototypes of the cryogenic 4–8-GHz isolator and low-noise amplifier used in the HIFI instrument (from
PamTech,5 and the Centro Astronomico de Yebes [23], respec5Passive
Microwave Technology Inc., Camarillo, CA.
552
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 2, FEBRUARY 2006
tively). (The input noise temperature of this IF system is 8 K,
which is only slightly less than the expected 10 K noise of the
IF chain in the HIFI instrument.) Furthermore, the SIS devices
have been tested in prototypes of the mixer blocks that will be
used in the instrument’s 0.88- and 1.04-THz bands (see [12]), at
an operating temperature of 2–2.5 K (which is close to the expected operating temperature of the mixers in the instrument).
(This low operating temperature is reached by pumping on the
cryostat’s helium bath.)
In reporting the receiver sensitivities presented here, two
, the measured double
values are generally given:
side-band (DSB) receiver sensitivity obtained with the mixer
mounted in the test receiver (using a power meter to detect the total output power over the 4–8-GHz IF band), and
, the effective input noise of the mixer and the IF
amplifier system (which is obtained by correcting the measured
receiver noise temperatures for the calculated/measured losses
in the 14- or 49- m Mylar beamsplitter and the Zitex G104
infrared heat-filters).
Finally, the mixers’ direct-detection sensitivities are also presented here because they provide a snapshot of the frequency
dependence of the coupling of radiation from an incident optical beam to the SIS junctions. These results have been obtained
with a Fourier transform spectrometer in which an evacuated
Michelson interferometer is used as a tunable signal source that
is injected into the mixer test cryostat via the optical window
through which the “hot” signal passes in heterodyne sensitivity
measurements. In order to allow the measured results to be compared with the calculated coupling efficiencies, the measured
spectra have been corrected for a standing wave in the output of
the Michelson interferometer (with a period of 70 GHz and a
peak-to-peak amplitude of 0.8 dB) and have been multiplied by
a factor of (in THz) to account for the fact that the measured
spectra are proportional to the photon detection efficiency, while
the calculated spectra are of the power coupling.
VI. RF MEASUREMENT RESULTS
Fig. 7 presents the direct-detection sensitivities of several SIS
devices in each of the three mixers discussed here (the 0.88-THz
mixer with moderate- and high- junctions and the 1.04-THz
mixer with moderatejunctions). From these results, it is
seen that both current densities yield efficient coupling to the
SIS junctions over broad RF bandwidths. Furthermore, provided
that the response is properly centered on the target band (i.e., by
properly matching the junction sizes to the tuning circuit geometries), efficient coupling can be obtained over the full 0.8–0.96and 0.96–1.12-THz bands.
Following the direct-detection sensitivities, Fig. 8 presents
measured heterodyne sensitivities of several of the SIS devices
whose direct-detection sensitivities are plotted in Fig. 7. From
these results, it is observed that the 0.88-THz mixers yield
K and
K, or better, across a
significant fraction of the 0.8–0.96-THz band. The 1.04-THz
K and
K, or
mixers yield
better, across at least 1.03–1.15 THz (for a device tuned to this
frequency range). Furthermore, a number of additional observations can be made: 1) the sensitivities of the 0.88-THz mixers
Fig. 7. Direct-detection sensitivities of several SIS devices mounted in the
= 2:4–2:5 mV. In
0.88-THz and 1.04-THz mixer blocks, biased at V
all cases, the embedding geometry is a half-height waveguide design with
the substrate suspended in the substrate channel. (a) Results for 0.88-THz
mixers with junction current densities of J = 6:5 kA/cm . (b) Results for
0.88-THz mixers with J = 13 kA/cm . (c) Results for 1.04-THz mixers with
J
= 6:5 kA/cm . In each case, strong coupling over a broad RF bandwidth
is obtained (although the center-frequency and the frequency dependence of
the response is dependent on the tuning circuit geometry and junction size).
The geometries of these devices are summarized in Table II, using the device
labels identified in the legend of each figure. Note that (c) is adapted from [11].
incorporating 6.5 and 13 kA/cm junctions are similar (given
the device-to-device variability in their sensitivities); 2) as is
discussed further in [11], the use of an NbTiN ground plane
deposited at 400 C yields sensitive SIS mixers for frequencies
up to at least 1.12 THz; and 3) the input noise temperatures
of the 1.04-THz mixers are 40%–50% higher than those of
the 0.88-THz mixers. Note that receiver sensitivities have not
been measured over the full bandwidths of the mxiers, due
to a lack of LO power in some frequency bands (especially
below 0.86 THz and between 1.04–1.14 THz). However, the
JACKSON et al.: LOW-NOISE 0.8–0.96- AND 0.96–1.12-THz SIS MIXERS FOR HERSCHEL SPACE OBSERVATORY
553
Fig. 8. Measured DSB receiver noise temperatures (T
) for several of the devices/mixers whose direct detection (FTS) sensitivities are plotted in Fig. 7.
At each frequency, the LO power level is adjusted to obtain optimum noise, while the bias voltage is held constant with frequency (at bias voltages that are indicated
) (obtained by correcting the measured receiver
in Fig. 5). The effective input noise temperatures of the combination of the mixers and IF system (T
sensitivities for the calculated/measured losses in the receiver optics), are also shown. All of these measurements are performed at mixer temperatures of 2–2.5 K,
averaging over the full bandwidth of a 4–8 GHz IF system with T
8–10 K. (a) Results for 0.88-THz mixers with J = 6:5 kA/cm . (b) Results for 0.88-THz
mixers with J = 13 kA/cm . (c) Results for 1.04-THz mixers with J = 6:5 kA/cm . In each case, high sensitivity over a broad RF bandwidth is obtained
(although the center frequency and the frequency dependence of the response is dependent on the tuning circuit geometry and junction size). The geometries of
these devices are summarized in Table II using the device labels indicated in the legends in the bottom right corner of each figure. Note that (c) is adapted from
[11]. Note that the LO frequency ranges over which sensitivities are shown are limited by the tuning ranges of the Backward Wave Oscillators that are available.
However, the direct detection spectral response curves of these devices, which are seen in Fig. 7, show that the coupling of incident radiation to the junctions over
the relevant 0.8–0.96- and 0.96–1.12-THz bands is reasonably good.
direct-detection spectral response curves in Fig. 7 show that
the coupling of incident radiation to the junctions remains
reasonably efficient over the majority of the relevant 0.8–0.96and 0.96–1.12-THz bands.
The determination of the DSB receiver conversion gains for
these mixers was complicated by parasitic impedances in the
mixers’ IF output circuit, which introduced uncertainties to the
use of the junction’s unpumped shot-noise characteristics to calibrate the noise and gain of the receiver’s IF system. However,
despite these uncertainties, clear differences could be seen in
the gains of the mixers from different wafers—c78-A offered
dB at 0.88 THz, c78-B offered
dB at 0.86 THz, and f49-C offered
dB at 1.04 THz.
The wider RF bandwidths that are expected from the use of
higher current-density junctions (see Fig. 4) are not immediately
obvious in the measured results in Figs. 7 and 8. However, this
may be partly due to the 1-THz “gap frequency” of the NbTiN
ground plane limiting the mixers’ high-frequency performance
(since the devices presented here have been selected to have
strong responses in the 0.8–0.96- or 0.96–1.12-THz bands, as
opposed to maximum bandwidth).
VII. DISCUSSION
A. Twin-Junction Mixer and Receiver Noise Model
Calculations of the coupling of radiation to the SIS junctions clearly show that the coupling efficiency increases significantly with increasing current-density (from 55% at
kA/cm to 70% at
kA/cm for the 0.88-THz mixer
design). However, measurements of the heterodyne sensitiviand 13 kA/cm show no signifties of devices with
icant difference in performance. This may be attributed to the
fact that the leakage currents of the higher current-density junctions are higher than those of the lower current-density juncin place of 30–60). In particular, the
tions (
larger leakage currents apparently generate sufficient additional
shot noise to counteract the improved coupling of radiation to
the junctions. (Note that the “sharpness” of the current–voltage
characteristics of the lower and higher current-density devices
in Fig. 5 is effectively the same, after correcting for the series
–
mV.)
resistance in their wiring layers—
This conclusion is supported by an analysis of the heterodyne
performance of the twin-junction tuning circuit. In this analysis, each of the microstrip transmission-line sections in the RF
554
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 2, FEBRUARY 2006
tuning circuit is replaced by a -matrix lumped-element circuit
model, and the coupling of LO power to the two junctions is
then determined in an iterative calculation that takes into account the dependence of the junction admittances on the absorbed LO powers. Given the (different) LO power coupling to
the two junctions, the junctions are then replaced by three-port
Tucker admittance and noise correlation matrices [24] in order
to generate a three-port model for the complete tuning circuit,
from which the frequency-dependent mixer noise and gain can
be determined. This model of the two-junction tuning circuit is
similar to those developed previously in [25] and [26].
In order to evaluate the impact of the junction quality and
current density on the mixer noise, the junction model is modified by adding a parallel resistance to the measured currentkA/cm and
voltage characteristic of a junction with
, and then scaling the voltage and current
to obtain the desired gap voltage and junction resistance. (Note
that adding this parallel resistance does not change the “sharpness” of the current step and the junction’s gap voltage, as is
the case in the mixers discussed here.) Finally, the shot noise
that is produced by this “excess leakage current” is multiplied
by the bias-voltage-dependent factor that was developed in [27]
and [28] to account for the amplification of shot noise by multiple Andreev reflection in the leakage current of “leaky” SIS
mV).
junctions (this factor is 2.3 at
Receiver noise temperatures are calculated from the calculated mixer noise temperatures and gains by adding input optical losses that are representative of those in the experimental
test setup that is being used, plus an IF system with an input
noise temperature of 10 K and a 2 K physical temperature. A
10 K IF system noise temperature is used because it is representative of the 4–8-GHz IF system in the HIFI instrument, in
which thermal and mechanical constraints have driven the use
of a relatively long stainless steel semi-rigid cable between the
mixer (at 2 K) and the cryoamplifier (at 15 K), plus two isolators
(one at each end of the long cable).
Fig. 9 presents a comparison of the measured and calculated
RF coupling, receiver noise, and receiver conversion gain for
two 0.88-THz devices with different junction resistances and
tuning circuit dimensions (devices c78-A and c87-A). Because
the tuning circuit dimensions and material properties are not
known with absolute certainty, the tuning circuit dimensions
have been fine-tuned to match the frequency dependence of
the calculated RF coupling efficiencies to the measured directdetection sensitivities (applying the same corrections to the
geometries of both devices). Furthermore, an additional lossy
element is added to the model to fit the calculated noise temperatures to the measured results. For the sake of argument,
this excess loss (and the noise that it generates) is represented
as a 2.8-dB loss in front of the mixer, at a physical temperature
of 2 K. As is seen in Fig. 9, the result is a reasonably consistent
match between the measured and calculated receiver noise
temperatures for the two mixers. The fact that the calculated
receiver conversion gain of c78-A is somewhat lower than the
measured gain indicates that at least part of the observed excess
receiver noise is likely due to “warm” losses. (Note that the
2.8 dB of excess loss at 2 K that is assumed in Fig. 9 can be
replaced by 1.1 dB of excess loss at 300 K.)
Fig. 9. Comparison of the measured and calculated direct-detection
sensitivities, receiver noise temperatures, and DSB receiver conversion
gains for devices c78-A and c87-A from the low-J 0.88-THz mixer (with
J
= 6:5 kA/cm and R
=R
= 60). Note that the tuning circuit
dimensions in the calculations have been tuned slightly to match the frequency
dependences of the measured direct-detection sensitivities. Matching the
calculated receiver noise temperatures to the measured values requires that an
excess loss be inserted in the model (i.e., 2.8 dB in front of the mixer, at a
physical temperature of 2 K). The receiver conversion gain values have been
corrected for the gain of the receiver’s IF system (the error bars in the measured
conversion gains stem from uncertainties in estimating this IF system gain).
The calculated excess noise/loss may originate from a
number of sources, including the 77 K blackbody load not
being perfectly black (although previous measurements show
0.95 [22]); excess losses
that it should have an emissivity of
in the receiver optics; losses in the corrugated horn, waveguide,
and/or substrate channel due to manufacturing errors and/or
resistive losses that are not included in the model; dielectric
losses in the fused quartz substrate and/or the SiO dielectric
layer in the tuning circuit, both of which are assumed to be
lossless; excess resistive losses in the Al wiring layer and/or
the NbTiN ground plane; resistive losses in the Nb junction
electrodes (which are not taken into account in this model, but
JACKSON et al.: LOW-NOISE 0.8–0.96- AND 0.96–1.12-THz SIS MIXERS FOR HERSCHEL SPACE OBSERVATORY
555
Fig. 10. Calculated sensitivities of 0.88-THz SIS mixers with twin-junction NbTiN/SiO /Al tuning circuits and junction current densities ranging from 6.5 to
15 kA/cm . The same noise and gain contributions for the receiver optics and IF system that were used in Fig. 9 are used in these calculations (including the 2.8-dB
excess loss in front of the mixer). (a) Frequency dependence of the direct-detection coupling to the SIS junctions for different current densities. (b) and (c) Frequency
dependence of the DSB receiver noise for different combinations of current density and junction quality (R
=R ). (d) Frequency dependence of the DSB
receiver noise for different current densities, with the junction quality being defined at each current density as the value that is needed to obtain the same sensitivity
as is obtained with J = 6:5 kA/cm and R
=R
= 60. A mixer bias voltage of 2.2 mV is used for all calculations, which is close to the optimum in all
cases, while the input noise of the IF system is assumed to be 10 K, which is representative of the 4–8-GHz IF system in the HIFI instrument. For each combination
of current density and junction quality, a frequency-independent LO power is used that optimizes the average noise temperature across the 0.8–0.96-THz band.
which are present); and impedance mismatches and/or resistive
losses in the mixer’s IF circuit (which is represented by a 50
load and an input noise of 10 K).
B. Junction Current Density and Quality Versus
Receiver Noise
Using the previously described receiver noise model of the
twin-junction mixer, including the 2.8-dB excess loss term that
was determined from the results in Fig. 9, the receiver noise of
the 0.88-THz mixer has been calculated as a function of junction current density and quality (while adjusting the junction
size and the dimensions of the tuning circuit’s transformer section to optimize the average RF coupling efficiency over the
0.8–0.96-THz band for each case). The results of these calculations are summarized in Fig. 10.
Note that, for each combination of current density and
junction quality, the LO power level has been adjusted to
optimize the average sensitivity across the band. (Within the
0.8–0.96-THz band, the sensitivities offered by these “averaged” optimum LO powers do not vary significantly from
those that are obtained if the LO power is optimized at each
frequency.) The optimum LO power thus increases with both
current density and leakage current, although the increase in
LO power with current density is driven by a corresponding
reduction in junction resistance—the pumping level of the
,
junction, as measured by the alpha parameter
does not change significantly with current density. A bias
voltage of 2.2 mV is used for all calculations—this value is
close to the optimum in all cases.
Not surprisingly, it is seen that, if the junction quality
remains constant, then the receiver noise temperature drops significantly with increasing current density
[see Fig. 10(b)]. However, experience shows that junction
quality drops with increasing current density in high current
–
kA/cm ) SIS junctions with AlO
density (
tunnel barriers. This reduction in junction quality causes a
corresponding increase in mixer shot noise [27], [28] that will
(partially) offset the improved coupling of incident radiation to
the junctions [see Fig. 10(c)]. Indeed, Fig. 10(d) presents the
calculated receiver sensitivities for current densities between
6.5 and 15 kA/cm , with a junction quality in each case that
is defined as the minimum value that is needed for junctions
with that current density to offer a receiver sensitivity that is
kA/cm and
equal to that which is obtained with
kA/cm .
Reviewing these calculations, it is noted that moving from
to 10 and 15 kA/cm requires that junction qualities of
and , respectively, are maintained
at least
556
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 2, FEBRUARY 2006
Fig. 11. Calculated receiver and mixer sensitivities (T
and T
)
for 0.88- and 1.04-THz mixers with current densities of J = 6:5 kA/cm
and junction qualities of R
=R
= 60. The lower sensitivity of the
1.04-THz mixers is due to a combination of the increased operating frequency
(which increases the noise contributions of RF coupling losses and reduces
the intrinsic conversion efficiency of the junctions) and the reduced coupling
efficiency of the tuning circuit (due to the use of a NbTiN ground plane with a
higher conductivity, and thus a smaller penetration depth). (Due to the reduced
penetration depth of the NbTiN ground plane deposited at 400 C, the microstrip
sections need to be made longer and narrower to keep the same electrical length
and impedance, which increases the series resistance in the Al wiring layer.)
For reference, indicators of the measured receiver sensitivities of the 0.88- and
1.04-THz mixers presented in Fig. 8 are also included.
in order to not lose receiver sensitivity. What is particularly interesting about this result is the fact that it is consistent with the
observation that no significant difference is seen in the heteroand 13 kA/cm and
dyne performances of mixers with
and , respectively. It is also interesting
to note that these calculations are also in qualitative agreement
with previous observations [29] that 0.64–0.8-THz mixers with
kA/cm and
yield similar sensitivities as mixers with
kA/cm and
.
Note that the increase in the mixers’ broad-band sensitivities
that is obtained with higher current-density junctions is recognized [see Fig. 10(d)]. However, for the purpose of this comparison, only the sensitivities in the 0.8–0.96-THz range are
considered.
C. Comparison of the 0.88- and 1.04-THz Results
This twin-junction mixer sensitivity model can also be used
to compare the relative sensitivities of the 0.88- and 1.04-THz
mixers described in this paper. In particular, Fig. 11 presents the
calculated sensitivities of 0.88- and 1.04-THz mixers in which
the tuning circuit and junction geometries have been fine-tuned
to center the mixers’ responses on their respective frequency
bands. From this plot, it is seen that a large fraction of the difference in the measured sensitivities of the two mixers is attributable to factors that are included in the previously described
mixer noise model. These factors include a drop in the junctions’
intrinsic conversion efficiencies and an increase in the noise contributions of optical losses with increasing operating frequency.
Additionally, because the penetration depth in NbTiN films deposited at 400 C is expected to be smaller than that of films
deposited at room temperature (due to their significantly higher
normal-state conductivities), the characteristic impedance of a
given width of NbTiN/SiO /Al microstrip transmission line is
reduced. As a result, maintaining the impedance and electrical
lengths of the microstrip sections in the tuning circuit requires
that they be made longer and narrower when replacing a roomtemperature-deposited film with one deposited at 400 C, which
increases the series resistance of the Al wiring layer. Finally, it
is noted that the 1.04-THz devices presented here are characterized by a series resistance in their current–voltage characteristics that is a factor of two higher than that in the 0.88-THz
devices. This is a sign that the low-temperature resistivity of
the Al wiring layer is likely higher in these devices than the
cm that is realized in the 0.88-THz devices.
If this factor-of-two increase in the resistivity of the Al wiring
layer is included in the noise model for the 1.04-THz mixer, the
calculated receiver noise increases by 10% across the band,
K that is obwhich brings it much closer to the
tained with device f49–C at 1.04 and 1.14 THz.
VIII. CONCLUSION
The development of SIS mixer devices incorporating
NbTiN/SiO /Al microstrip tuning circuits and a parallel pair of
“standard” 1– m Nb/Al-AlO /Nb SIS junctions has enabled
the development of low-noise mixers in the 0.8–1.12-THz
range. Furthermore, the use of half-height 0.88- and 1.04-THz
waveguide embedding geometries and the optimization of SIS
devices to couple efficiently to these embedding circuits has
yielded low receiver noise temperatures across the 0.8–0.96and 0.96–1.12-THz bands of the HIFI instrument; a high-resolution heterodyne spectrometer that is being built for the
European Space Agency’s Herschel Space Observatory.
Analyzing the measured mixer performance with a three-port
admittance and noise model of the SIS tuning circuit (in which
three-port “Tucker” models of the SIS junctions are combined
with lumped-element representations of the microstrip transmission lines), it is found that the frequency dependence of the
measured performance is consistent with calculations, but that
the measured noise can only be reproduced by inserting 2.8 dB
of excess loss into the noise model (in front of the mixer, at
a physical temperature of 2 K). The calculated results are also
consistent with the observation that mixers containing junctions
with current densities of
and
kA/cm yield similar sensitivities. This may be attributed to the fact that the improved RF coupling that is obtained with higher current-density junctions is offset by an increase in junction shot-noise (due
to the fact that the higher current-density junctions have larger
leakage currents—
for
kA/cm
versus
– for
kA/cm ).
Finally, a significant fraction of the drop in sensitivity of the
1.04-THz devices relative to the 0.88-THz devices is attributed
to the intrinsic effects of the higher operating frequency (which
causes the conversion gains of the SIS junctions to drop and
the noise contributions of RF losses to increase) and the smaller
penetration depth in the “higher quality” NbTiN films used in
these devices.
JACKSON et al.: LOW-NOISE 0.8–0.96- AND 0.96–1.12-THz SIS MIXERS FOR HERSCHEL SPACE OBSERVATORY
ACKNOWLEDGMENT
The authors would like to thank W. Laauwen and L. de Jong
for performing the mixer measurements described here,
M. Eggens, H. Golstein, S. Kikken, D. Nguyen, C. Pieters,
H. Schaeffer, and H. Smit for their contributions to the design
and construction of the mixers and test systems that were
used in this work, and A. Baryshev, J. R. Gao, T. de Graauw,
N. Honingh, R. LeDuc, B. Leone, S. Shitov, N. Whyborn, and
J. Zmuidzinas for useful discussions.
REFERENCES
[1] T. de Graauw and F. P. Helmich, “Herschel-HIFI: The heterodyne instrument for the far infrared,” in SP-460 The Promise of the Herschel Space
Observatory, G. L. Pilbratt, J. Cernicharo, A. M. Heras, T. Prusti, and
R. A Harris, Eds. Noordwijk, The Netherlands: ESA Pub. Div., 2001,
pp. 45–52.
[2] G. L. Pilbratt, J. Cernicharo, A. M. Heras, T. Prusti, and R. A Harris,
SP-460 The Promise of the Herschel Space Observatory. Noordwijk,
The Netherlands: ESA Pub. Div., 2001.
[3] J. W. Kooi, M. Chan, B. Bumble, H. G. LeDuc, P. Schaffer, and T. G.
Phillips, “230 and 492 GHz low noise SIS waveguide receivers employing tuned Nb/AlO /Nb tunnel junctions,” Int. J. Inf. Millim. Waves,
vol. 16, pp. 2049–2068, Dec. 1995.
[4] A. Karpov, J. Blondel, M. Voss, and K. H. Gundlach, “A three photon
noise SIS heterodyne receiver at submillimeter wavelength,” IEEE
Trans. Appl. Supercond., vol. 9, no. 6, pp. 4456–4459, Jun. 1999.
[5] C. E. Honingh, S. Haas, D. Hottgenroth, K. Jacobs, and J. Stutzki,
“Low noise broadband fixed tuned SIS waveguide mixers at 660 and
800 GHz,” IEEE Trans. Appl. Supercond., vol. 7, no. 6, pp. 2582–2586,
Jun. 1997.
[6] D. C. Mattis and J. Bardeen, “Theory of the anomalous skin effect in
normal and superconducting metals,” Phys. Rev., vol. 111, pp. 412–417,
1958.
[7] G. de Lange, J. J. Kuipers, T. M. Klapwijk, R. A. Panhuyzen, H. van
de Stadt, and M. W. M. de Graauw, “Superconducting resonator circuits at frequencies above the gap frequency,” J. Appl. Phys., vol. 77,
pp. 1795–1804, Feb. 1995.
[8] J. Kawamura, J. Chen, D. Miller, J. Kooi, J. Zmuidzinas, B. Bumble,
H. G. Leduc, and J. A. Stern, “Low-noise submillimeter-wave NbTiN
superconducting tunnel junction mixers,” Appl. Phys. Lett., vol. 75, pp.
4013–4015, Dec. 1999.
[9] B. D. Jackson, A. M. Baryshev, G. de Lange, S. V. Shitov, J.-R. Gao,
N. N. Iosad, and T. M. Klapwijk, “Low-noise 1 THz superconductorinsulator-superconductor mixer incorporating a NbTiN/SiO /Al tuning
circuit,” Appl. Phys. Lett., vol. 79, pp. 436–438, Jul. 2001.
[10] B. D. Jackson, N. N. Iosad, G. de Lange, A. M. Baryshev, W. M.
Laauwen, J.-R. Gao, and T. M. Klapwijk, “NbTiN/SiO /Al tuning circuits for low-noise 1 THz SIS mixers,” IEEE Trans. Appl. Supercond.,
vol. 11, no. 3, pp. 653–656, Mar. 2001.
[11] B. D. Jackson, G. de Lange, T. Zijlstra, M. Kroug, T. M. Klapwijk, and
J. A. Stern, “Niobium titanium nitride based superconductor-insulatorsuperconductor mixers for low-noise THz receivers,” J. Appl. Phys., vol.
97, no. 11, p. 113 904 1–8, Jun. 2005.
[12] G. de Lange, B. Jackson, T. Zijlstra, M. Kroug, and T. M. Klapwijk, “Development of the band 3 and 4 mixer units for HIFI,” in Millimeter and
Submillimeter Detectors for Astronomy, Proc. of the SPIE, J. Zmuidzinas, W. S. Holland, and S. Withington, Eds. Bellingham, WA: SPIE,
2004, vol. 5498, pp. 268–277.
[13] J. W. Kooi, private communication, 2000–2003.
[14] B. D. Jackson, G. de Lange, W. M. Laauwen, L. de Jong, T. Zijlstra, N.
N. Iosad, and T. M. Klapwijk, “THz SIS mixer development for HIFI,”
in Proc. 13th Int. Symp. Space THz Technol., R. Blundell and E. Tong,
Eds. Cambridge, MA, 2002, pp. 561–570.
[15] H. van de Stadt, H. Scheaffer, and L. de Jong, private communication,
1996–1998.
557
[16] A. M. Baryshev, H. van de Stadt, H. Schaeffer, R. Hesper, T. Zijlstra,
M. Zuiddam, W. Wild, and L. de Jong, “Development of a 0.6 THz SIS
receiver for ALMA,” in Proc. 12th Int. Symp. Space THz Technol., I.
Mehdi, Ed.. San Diego, CA, 2001, pp. 581–590.
[17] R. L. Eisenhart and P. J. Khan, “Theoretical and experimental analysis
of a waveguide mounting structure,” IEEE Trans. Microw. Theory Tech.,
vol. MTT-19, no. 8, pp. 706–719, Aug. 1971.
[18] C.-Y. E. Tong, R. Blundell, S. Paine, D. C. Papa, J. Kawamura, X.
Zhang, J. A. Stern, and H. G. LeDuc, “Design and characterization
of a 250–350-GHz fixed-tuned superconductor-insulator-superconductor receiver,” IEEE Trans. Microw. Theory Tech., vol. 44, no. 9, pp.
1548–1556, Sep. 1996.
[19] J. A. Stern, B. Bumble, H. G. Leduc, J. W. Kooi, and J. Zmuidzinas,
“Fabrication and dc-characterization of NbTiN based SIS mixers for use
between 600 and 1200 GHz,” in Proc. 9th Int. Symp. Space THz Technol.,
R. McGrath, Ed.. Pasadena, CA, 1998, pp. 305–313.
[20] H. B. Callen and T. A. Welton, “Irreversibility and generalized noise,”
Phys. Rev., vol. 83, pp. 34–40, 1951.
[21] T. O. Klaassen, M. C. Diez, J. H. Blok, C. Smorenburg, K. J. Wildeman,
and G. Jakob, “Optical characterization of absorbing coatings for submillimeter radiation,” in Proc. 12th Int. Symp. Space THz Technol., I.
Mehdi, Ed., San Diego, CA, 2001, pp. 400–409.
[22] N. D. Whyborn, private communication, Jul. 2005.
[23] I. Lopez-Fernandez, J. D. G. Puyol, A. B. Cancio, and F. Colomer, “New
trends in cryogenic HEMT amplifiers for radio astronomy,” presented at
the Int. Science and Technology Meeting on the Square Kilometer Array,
Berkeley, CA, Jul. 9–13, 2001.
[24] J. R. Tucker and M. J. Feldman, “Quantum detection at millimeter wavelengths,” Rev. Mod. Phys., vol. 57, pp. 1055–1113, Oct. 1985.
[25] J. Zmuidzinas, H. G. Leduc, J. A. Stern, and S. R. Cypher, “Two-junction tuning circuits for submillimeter SIS mixers,” IEEE Trans. Microw.
Theory Tech., vol. 42, no. 4, pp. 698–706, Apr. 1994.
[26] T. Noguchi, S. C. Shi, and J. Inatani, “Parallel connected twin SIS junctions for millimeter and submillimeter-wave mixers—Analysis and experimental verification,” IEICE Trans. Elect., vol. E78C, pp. 481–489,
May 1995.
[27] P. Dieleman, H. G. Bukkems, T. M. Klapwijk, M. Schicke, and K. H.
Gundlach, “Observation of Andreev reflection enhanced shot noise,”
Phys. Rev. Lett., vol. 79, pp. 3486–3489, Nov. 1997.
[28] P. Dieleman and T. M. Klapwijk, “Shot noise beyond the Tucker theory
in niobium tunnel junction mixers,” Appl. Phys. Lett., vol. 72, pp.
1653–1655, Mar. 1998.
[29] R. Teipen, M. Justen, T. Tils, S. Glenz, C. E. Honingh, K. Jacobs, B. D.
Jackson, T. Zijlstra, and M. Kroug, “Influence of junction-quality and
current density on HIFI band 2 mixer performance,” in Proc. 14th Int.
Symp. Space THz Technol., C. Walker and J. Payne, Eds., Tucson, AZ,
2003, pp. 55–62.
Brian D. Jackson received the B.A.Sc. degree in
engineering physics from the University of British
Columbia, Vancouver, BC, Canada, in 1995, the
M.S. degree in electrical and computer engineering
from the University of Toronto, Toronto, ON,
Canada, in 1997, and the Ph.D. degree in applied
physics from the Delft University of Technology,
Delft, The Netherlands, in 2005.
He was a Research Assistant with the National Research Council of Canada, Ottawa, ON, Canada, in
1995, and with the University of Toronto from 1995
to 1997. From 1997 to 1999, he was a Research Engineer with the University of
Groningen, Groningen, The Netherlands. Since 1999, he has been an Instrument
Scientist with the SRON Netherlands Institute for Space Research, Groningen,
The Netherlands. His main research activities are in the areas of SIS junction
and mixer development for radio astronomy and system engineering for heterodyne instruments for space- and ground-based telescopes. He has played a
role in SIS mixer and/or receiver development for the James Clerk Maxwell
Telescope, the Heterodyne Instrument for the Far-Infrared (HIFI) on the European Space Agency’s Herschel Space Observatory, and for the Atacama Large
Millimeter Array.
558
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 54, NO. 2, FEBRUARY 2006
Gert de Lange received the M.S. and Ph.D. degree
from the University of Groningen, Groningen,
The Netherlands, in 1988 and 1994, respectively,
both in applied physics.
From 1994 to 1998, he was a Research Scientist
with the Massachussetts Institute of Technology,
Cambridge, working on the development of superconducting micromachined heterodyne detector
arrays. Since 1998, he has been an Instrument
Scientist with the SRON Netherlands Institute
for Space Research, Groningen, The Netherlands.
His main research activities are in the areas superconducting detector development for low-noise heterodyne receivers for space- or ground-based and
airborne submillimeter astronomical instrumentation and the development of
space qualified mixer units for the Heterodyne Instrument for the Far-Infrared
(HIFI) on the European Space Agency’s Herschel Space Observatory.
Tony Zijlstra, photograph and biography not available at the time of
publication.
Matthias Kroug, photograph and biography not available at the time of
publication.
Jacob W. Kooi, photograph and biography not available at the time of
publication.
Jeffrey A. Stern received the B.S. degree in physics
from Rensselaer Polytechnic Institute, Troy, NY, in
1983, and the Ph.D. degree in applied physics from
the California Institute of Technology, Pasadena, in
1991.
Since receiving his doctorate, he has been with the
Jet Propulsion Laboratory, Pasadena, most recently
as a Senior Member of the Technical Staff. His work
has focused on a number of superconducting sensors
including Nb NbTiN and NbN SIS mixers, NbN
and NbTiN phononcooled, hot-electron mixers, and,
more recently, NbN single0photon detectors. He was in charge of fabricating,
space qualifying, and delivering mixer chips for band 5 of the HIFI instrument
on Hershel. He has also been involved in delivering SIS mixer chips to the
Caltech Millimeter Array and the Harvard Smithsonian Submillimeter Array.
He has also delivered 1.4 THz NbTiN HEB mixers for the Harvard Smithsonian
Receiver Lab Telescope in Chile.
Teun M. Klapwijk received the Ph.D. degree in
applied physics from the Delft University of Technology, Delft, The Netherlands, in 1977. His doctoral
degree was entitled “Superconducting Microbridges
and Radiation Stimulated Superconductivity.”
After receiving the doctorate degree, he continued
his research with Delft University as an Assistant/Associate Professor, interrupted by periods
as a Research Fellow with Harvard University
(1979–1980) and as a Summer-Faculty member
with the IBM T. J. Watson Research Laboratory,
Yorktown Heights, NY (1983). In 1985, he became a Full Professor with
the University of Groningen, where he was involved with silicon MOSFETs,
superconductor/semiconductor hybrids, and began his long-term collaboration
with the SRON Netherlands Institute of Space Research on heterodyne detection with superconducting tunnel junctions and hot-electron bolometers. This
research has found its application at the James Clerk Maxwell Telescope, for the
Herschel Space Telescope and the Atacama Large Millimeter Array. In 1999,
he returned to Delft University, transferring a portion of his ongoing research
program and moving into new areas such a superconducting/ferromagnetic
hybrids. He is currently a Professor of applied physics, specializing in the field
of nanoelectronics, with the Faculty of Applied Sciences, Kavli Institute of
Nanoscience, Delft University of Technology, Delft, The Netherlands. He has
authored or coauthored over 250 refereed scientific publications.
Prof. Klapwijk has been an elected Fellow of the American Physical Society
since 2001.